Phishing Goes Supersonic With AI

Illustration of a globe with digital connections and blue lighting

AI is now helping criminals run more of a cyberattack, from scouting targets to fooling victims and pushing malware.

Quick Take

Check Point says AI is now used across multiple attack stages, not just for simple support tasks.
The company reported an average of 1,968 cyber attacks a week in 2025, up 70 percent since 2023.
Check Point also found that 89 percent of organizations faced risky AI prompts in a three-month period.
Other research says AI makes phishing, recon, and malware work faster and harder to spot.

AI Is Moving Into the Attack Chain

Check Point’s 2026 cyber security report says attackers are using AI across the attack lifecycle, including reconnaissance, social engineering, malware development, and data exploitation. The report argues that this shift makes attacks faster, more adaptive, and more convincing. That matters because cybercrime is no longer limited to skilled operators working by hand. AI now helps criminals scale work that once took time, effort, and a clear trail.

Check Point also said organizations faced an average of 1,968 attacks per week in 2025, which the company called a 70 percent jump from 2023. The report links that rise to more automation in the hands of attackers. It also says 89 percent of organizations encountered risky AI prompts in a three-month window, and one in 41 prompts was judged high risk. Those numbers show how fast the threat surface is widening.

Why Social Engineering Is a Big Concern

Several sources say AI is strongest where human judgment is weakest. The National Academies workshop on artificial intelligence and cybersecurity said AI can use social media and text generation to make spearphishing messages much harder to tell from normal email. CrowdStrike added that AI can identify targets, build personas, write tailored messages, and even automate live phishing chat in advanced cases. That gives criminals a cheap way to pressure families, workers, and businesses.

PwC said AI is being used across multiple stages of the attack lifecycle and can create highly convincing spearphishing and smishing campaigns much faster than humans can. The same report said AI removes many of the old warning signs that helped people spot scams. For readers who already distrust digital manipulation, that is the most troubling part. The fraud is becoming cleaner, faster, and harder to catch before damage is done.

What the Evidence Shows, and What It Does Not

The broader research supports the core trend: AI is helping attackers with reconnaissance, phishing, malware work, and target selection. At the same time, the material provided here does not prove that every claimed AI attack is fully autonomous. The strongest public evidence shows heavy AI use and reduced human effort, not a complete end to human control. That distinction matters for accuracy, even if the threat is still serious.

Discover how autonomous AI cyber attacks orchestrate entire hacks. Read the latest Check Point research detailing this dangerous shift in network security.#Cybersecurity #ArtificialIntelligence #CheckPoint #CyberAttacks https://t.co/QyuZMlf7Kd

— Daily CyberSecurity (@Daily_CyberSec) July 16, 2026

One counterpoint in the research is that some AI attacks still need people at key decision points. That does not erase the danger. It does show that today’s threat is often a hybrid model, where AI does much of the work and a human still steers the campaign. For defenders, that means old assumptions are failing. A criminal does not need full machine independence to cause major harm.

What This Means for Defenders

The practical lesson is simple. AI is lowering the cost of cybercrime and raising the speed of attacks. Check Point, CrowdStrike, PwC, and the National Academies all point to the same direction: better targeting, faster phishing, and more efficient abuse of public data. That should push companies to tighten training, watch for unusual prompts, and treat synthetic messages as a real threat, not a future theory.